The formylmethanofuran:tetrahydromethanopterin formyltransferase from Methanobacterium thermoautotrophicum delta H. Nucleotide sequence and functional expression of the cloned gene.

The formylmethanofuran:tetrahydromethanopterin formyltransferase (FTR) from Methanobacterium thermoautotrophicum delta H was cloned and its sequence was determined. The clone was contained on a 4.8-kilobase BamHI fragment of M. thermoautotrophicum DNA ligated into pBR329. When this fragment was subcloned into the phagemid pTZ18R, a functional enzyme was synthesized under control of the lac promoter. Sequence analysis revealed the presence of a ribosome binding site and a possible terminator structure. The absence of an identifiable promoter lends credibility to the open reading frame which is present 5' to ftr. The ftr gene encodes an acidic protein with a calculated molecular weight of 31,401. The sequence of FTR does not appear to be homologous to any other sequenced proteins, including proteins which use pterin substrates.

AH obligately reduces COz with hydrogen to yield methane. Despite the existence of gaps in the pathway, several intermediate conversions of methanogenesis have been well studied in this organism, and some of the enzymes have been purified (l-5). Many of these enzymes are unique to methanogens in that they catalyze interconversions among novel coenzymes (6). One of these enzymes, the formylmethanofuran:tetrahydromethanopterin (FTR),' has been demonstrated as an essential enzyme in the biosynthesis of methane (1). FTR transfers the formyl group of formylmethanofuran to tetrahydromethanopterin to yield 5-formyltetrahydromethanopterin (1). This enzyme differs remarkably from nonmethanogenic formylp- terin (e.g. lo-formyltetrahydrofolate) synthesizing enzymes (EC 6.3.4.3) which catalyze the ATP-dependent formation of a lo-formylated product (7). The biosynthesis of 5-formyltetrahydrofolate in nonmethanogenic bacteria is not clearly defined, however its conversion to the lo-formyl derivative is well characterized and proceeds in a two-step ATP-dependent reaction via 5,10-methenyltetrahydrofolate (8). Regardless of the mechanistic differences, the dependence of FTR on pterins justifies an inherent relatedness to other pterin-utilizing enzymes. In addition, the enzymes which catalyze the subsequent reactions, 5,10-methenyltetrahydromethanopterin cyclohydrolase (2) and 5,10-methylene-tetrahydromethanopterin dehydrogenase (3), can be considered analogs of the respective tetrahydrofolate enzymes of nonmethanogenic C-l biosynthesis despite catalytic differences.
The study of methanogenesis has proceeded entirely through biochemical approaches. Recently, some of the enzymes involved in this pathway have yielded to molecular characterization (for a review see Ref. 9). To characterize FTR at the molecular level, we have cloned and sequenced its gene. We found that when expressed in Escherichiu coli, the enzyme was catalytically active despite the thermophilic nature of its natural host (M. thermoautotrophicum AH grows optimally at 65-70 "C (10)). The methylcoenzyme M reductase from thermophilic and mesophilic methanogens has been cloned and expressed in E. coli (ll-13), and in both cases the products were nonfunctional; inactivity was due to the requirement of the novel prosthetic group, Factor 430. The functional expression of other methanogenic enzymes such as the methyl viologen reducing hydrogenase from M. thermoautotrophicum may be impeded by inherent oxygen lability (14). FTR, an oxygen stable enzyme with no unusual cofactor requirement, represents the first enzyme from the methane biosynthetic pathway to be functionally expressed in E. coli.

RESULTS AND DISCUSSION
The characterization of genes involved in the biogenesis of methane has been hindered by their indispensability in organisms like M. thermoautotrophicum which have no alternative metabolic pathway. In addition, the thermophilic and anaerobic nature of this organism necessitates the use of special culture conditions, thereby aggravating manipulative techniques. To circumvent the standard genetic strategies, we have decided to use an approach which takes advantage of the relatively large amount of biochemical information available, and clone the gene which encodes the formylmethanofuran:tetrahydromethanopterin formyltransferase. Genomic digests of M, thermoautotrophicum DNA were probed with the partially mixed pool of oligonucleotides described under "Experimental Procedures." Each digest yielded two hybridizing bands of unequal intensity. The sizes of the hybridizing bands were as follows: BamHI, 4.44 and 4.81 kb; EcoRI, 5.07 and 3.45 kb; P&I, 4.81 and 1.84 kb; with the stronger of the two bands listed first. The BamHI fragments were pursued by isolating BamHI-digested DNA in the size range of 4-5 kb, ligating into pBR329, and transforming E. coli HBlOl with the recombinant plasmid. Clones containing the two hybridizing fragments were identified and designated ADM5 (pADM5 contained the 4.8-kb fragment), and ADM7 (pADM7 contained the 4.4-kb fragment). These fragments were then subcloned into the phagemid pTZ18R, and the region hybridizing to the oligonucleotides was sequenced. The weaker hybridizing fragment (pADM5) was found to contain a nucleotide sequence which corresponded to the determined aminoterminal amino acid sequence of 60 residues. The recombinant pTZ18R derived phagemid containing the 4.8-kb fragment (pADM97) was used for subsequent analyses. The clone ADM7 was sequenced and shown to contain a sequence which matched identically with the first 13 nucleotides of the 14-nucleotide oligomer probe, however the flanking region did not agree with the determined amino acid sequence. The absence of additional hybridizing bands in the genomic digests is convincing evidence that the ftr gene is present as a single copy on the M. thermoautotrophicum chromosome.
Structure and Sequence of ftr-A physical map of the 4.8kb fragment is shown in Fig. 1 acidic residues which resulted in a net charge of -17.7 at pH 7, and an overall p1 4.48; highly charged peptides have been found to bind far less SDS per g of protein than the calibration proteins (24). This is supported by examples of anomalous behavior of highly acidic proteins on SDS-PAGE systems.
Acyl carrier protein from E. coli (p1 4.1) migrates with a molecular weight 20,000 protein while its actual molecular weight is 8847 (25). Formate dehydrogenase from Methunobacterium formicicum consists of two subunits with molecular weights determined from SDS-PAGE as 85,000 and 53,000.
The codon usage frequencies of ftr and the genes encoding the methylcoenzyme M reductase subunits (01, ,f3, y) are presented in Table I. With a few exceptions (Lys, Asp, and Glu), there is fairly good agreement.
Analysis of the noncoding regions 5' and 3' to ftr was based on comparison with methanogenic sequences (9, 11). The coding region of ftr is followed immediately by a region containing an inverted repeat sequence juxtaposed with a sequence of poly(dT).
This structure is suspected of serving as a transcriptional termination signal (9). A putative ribosome binding site exists at positions -11 to -6 in the 5' noncoding region. This sequence, (5'-AGGTGA-3'), is similar to the purine-rich sequences which precede the initiation codon of other genes from this organism (9, 11). There were no sequences in this upstream region which resembled a promoter sequence from either E. coli (27) or M. thermoautotrophicum (9, ll), however there exists a coding-like sequence in the region upstream of ftr which terminates at position -61 and is indicated in Fig. 2. This open reading frame could possibly encode the COOH terminus of a gene which precedes ftr as part of an operon. Support for this hypothesis is based on the existing pattern of gene arrangements studied in methanogenic bacteria which display a motif of operonic clustering (9). Investigation of the upstream open reading frame will provide information regarding this question.
Expression of the ftr Gene Product in E. coli-To express the ftr gene product in E. coli, a 2.5-kb BamHI/SalI fragment containing ftr was ligated into pTZ18R. This construction, shown in Fig. 3, allowed the ftr gene to follow the lac promotor, interrupted by the N-terminal portion of 1ucZ' and 211 bases of M. thermouutotrophicum DNA.
The resulting clone, ADMll, was grown under kc-inducing (+IPTG) and lucrepressing (+glucose) conditions and was subjected to electrophoresis and immunoblot analysis (Fig. 4). Although the luc-  (Fig. 4, lane G) with a mobility that coincided with homogeneous FTR from M. thermoautotrophicum, indicating that initiation of translation occurs at the same site. Were translation to start at either of the nearest in-frame initiation codons, the resulting polypeptides would differ by 2.8 kDa (initiation at position -85) or 7.8 kDa (initiation at position 220). While the accuracy of SDS-PAGE systems for the determination of molecular weights can be variable, the fact that the cloned ftr product and purified FTR from M. thermoautotrophicum have identical mobilities suggests that they are identical.
Induction of the lac promotor appears to be essential for the synthesis of FTR in the pADMl1 construct since no immunocross-reacting band is visible in extracts from cells grown under conditions where lac is repressed (Fig.   4, lane H), or where lac does not precede ftr (Fig. 4, lane E).
To test if the expressed FTR was active, ADMll was grown under &-inducing conditions (+IPTG), and the extract was assayed as described by Donnelly and Wolfe (1) at 60 "C. ADMll cell lysates exhibited a specific activity which was 3fold higher than the activity from M. thermoautotrophicum. However, when ADMll was grown under lac repressing conditions (-IPTG, +glucose), no FTR activity was detected. The same is true for ADM97 extracts grown under inducing conditions (pADM97 contains ftr in the opposite orientation with respect to lac). This clearly demonstrates that ftr transcription is entirely dependent on the lac promotor in the construct pADM11.
The functional expression of this thermostable enzyme in cells grown aerobically at 37 "C is an interesting phenomenon which has been demonstrated previously with other enzymes from both eubacterial (28) and archaebacterial (29, 30) sources. The implication that FTR is folded correctly in a foreign host when grown at a much lower temperature than in its natural host is especially encouraging for future work in cloning additional thermostable enzymes involved in methanogenesis. Comparison of FTR with Folate Binding Proteins-FTR represents a novel enzyme with no true catalytic homolog. To see if FTR shared structural homology with other proteins, a search of the Genbank'" protein sequence data bank was conducted. In addition, the protein sequences of formyltetrahydrofolate synthetase from a number of sources (31-33) were aligned with FTR by the method of Needleman and Wunsch (34) using the Aalign program of DNASTAR'" (Madison, WI). No extensive homologies were found. Despite the absence of extensive homology, it is possible that these or other folate utilizing proteins might share a common pterin binding sequence with FTR. To test this, protein sequences from thymidylate synthetase (5,10-methylenetetrahydrofolate:dUMP C-methyltransferase, EC 2.1.1.45) (35, 36), and dihydrofolate reductase (EC 1.5.1.3) (37, 38) were compared with FTR as described above. No common sequences were observed which could be designated as possible pterin binding sites. This is not surprising in light of the absence of common folate binding structures among different folate utilizing enzymes (32, 39). It appears that FTR has a novel pterin binding structure; whether or not this will be conserved among other tetrahydromethanopterin enzymes is unknown.